ChatGPT Ads · E-commerce and online stores
ChatGPT Ads for Online Stores
Buying decisions often start in chat as an advisory conversation: someone describes a situation, a budget or a problem and has the options sorted for them. For stores this is an early moment, before any brand is locked in. Your ad sits below the answer when your assortment matches the topic.
Online stores are not named explicitly on OpenAI's starting list, but product feed campaigns are documented: approval depends on your assortment, with home and lifestyle cleared first (OpenAI ad policies, as of 31 August 2026).
Conversation starters
How people ask in ChatGPT
Not search terms, conversations. This is roughly how the questions sound next to whose answer your ad should appear.
What gift works for someone who already has everything, up to fifty euros?
I need a mattress for back sleepers, what should I pay attention to?
What is the difference between these two models and is the pricier one worth it?
Where can I still get this delivered in time for the holidays?
How strict are return conditions with online stores actually?
Context hints
Context hints that match
Context hints describe the situation in which your offer helps. OpenAI matches them against the running conversation, not against a keyword.
People looking for a gift with a fixed budget who need delivery by a set date
First-time buyers who want to understand material differences before choosing a model
Households replacing a broken appliance and looking for one with the same fitting dimensions
People with allergies looking for products without certain ingredients who read the label first
Repeat customers restocking a consumable who pay attention to shipping costs
Examples following the OpenAI best practice, not exact match keywords.
Setup
What the campaign starts with
Campaign objective
For stores, order_created is the target event, with amount and currency so revenue shows up in reporting. Still start with clicks (CPC) and switch only once the event fires regularly; for broad assortments product feed campaigns are documented as an additional route.
Landing page
The ad points at the product or category page that matches the described need, with price, delivery time and return conditions visible without extra clicks. OAI-AdsBot and OAI-SearchBot must not be blocked, which is worth checking regularly on shop systems with strict bot defences.
From my practice
Three mistakes that tend to cost money
Setting up the product feed once and leaving it. From experience in auction channels, revenue leaks exactly where availability and price in the feed no longer match reality.
Counting on cart abandoners. OpenAI does not support custom audiences for EEA and Switzerland targeting, so this familiar lever is missing entirely.
Looking only at the last click. From my practice with auction channels, transferred to ChatGPT Ads: a conversation sits earlier in the decision, so checkout_started belongs in reporting next to order_created.
Pricing
What the start costs
Account, pixel, context hint map and the first campaign, live in 7 days.
3 months start phase, cancellable monthly after that.
Review of an existing account, including a list of actions.
Frequently asked questions
Are online stores eligible for ChatGPT Ads at all?
OpenAI's starting list names categories, not business models: home and lifestyle are explicitly included, and product feed campaigns are documented. Whether your shop runs depends on your assortment, and I clarify that before the first euro of budget.
Do I need a product feed or are individual ads enough?
Both are possible. Individual ads suit a small set of clearly describable bestsellers, a feed suits a broad assortment. From experience in auction channels, a feed only pays off once prices and availability stay current automatically.
How do I measure revenue and not just orders?
The standard event order_created accepts amount and currency, and currency is mandatory as soon as amount is set. Pixel and conversions API can be combined with the same event_id so an order is not counted twice.
Do I have to shift budget out of Google Shopping for this?
No. I run ChatGPT Ads as a test on its own budget next to the existing base, because OpenAI itself publishes no industry benchmarks yet. What works gets scaled, not the other way around.